Skip to content

platform capability

An AI knowledge base that answers
with facts, not fiction.

Services, prices, rules, documents and stop-topics in one knowledge base. The AI answers from your business facts, updates without a developer and hands the hard cases to a human.

3 days free on any plan. No card required.

live

A bad answer fixed in 30 seconds

A test dialogue shows how the AI answers from the knowledge base. Spot an error: explain the problem, and the instruction updates without a developer.

what it is

What is an AI knowledge base in Botseller?

An AI knowledge base in Botseller is a store of business facts an AI teammate answers clients from. Into it go services, prices, rules, documents, limits and stop-topics, and the AI uses exactly these materials, not a neural network guesswork: if the answer is not in the base or the question is sensitive, the dialogue is handed to a human with a summary and history. That removes the main fear of bots — a wrong fact is worse for a client than silence, especially in medicine, education and finance. The knowledge updates without a developer: conditions or the schedule changed, you edit the records in your own words, and the AI answers the new way across every channel — WhatsApp, Telegram, the site, voice. A wrong answer is fixed through debugging: you explain the problem on a test dialogue, the instruction updates, and the fix is checked before clients see it. The base is tied to the CRM, so the AI answers to the person actual situation.

how it works

01

Gather the knowledge

Services, prices, rules, documents, limits, stop-topics and the tone of voice: in your own words.

02

The AI answers by the rules

It does not invent, it uses the knowledge base, the client card and the allowed actions.

03

Hard cases go to a human

A sensitive or non-standard question the AI hands to an operator with a reason and a summary.

04

The knowledge improves

A wrong answer is fixed through debugging on a test dialogue, without rewriting the prompt.

why it matters

Why the AI knowledge base matters more than the neural network itself

Plugging a neural network into a chat is easy today, and it answers smoothly: politely, at length, without pauses. But a smooth answer is not a correct one. A general model is trained on the whole internet and knows neither your services nor your rules. Ask it about delivery times and it will confidently name a plausible number that appears in none of your documents.

For a business that confidence costs dearly. The client remembers the promise the bot gave and comes to a live admin to collect it. Sorting out each case falls to the owner, and every invented answer chips away at trust in the automation. A few such stories and the team switches the bot off to be safe: the experiment is over, everything is back on people.

So the AI knowledge base decides more than the choice of model. It sets the borders: here are the services, here are the terms, here are the topics the AI does not enter. Inside those borders the bot answers as confidently and precisely as your best admin, and beyond them it honestly calls a human. You set the rules, in plain text, and can change them any time. It even doubles as a self service knowledge base clients get straight answers from.

how it compares

How it differs from a rigid-script chatbot

By the job it is a familiar support bot, and by build a knowledge base AI: it answers repeat questions and takes load off admins and operators.

It understands a question asked in the person own words and answers from the knowledge base, instead of walking a button tree.

It knows its limits: sensitive and non-standard questions go to a human with a summary, not into a dead end.

It updates without a developer: conditions changed, you edit a record, you do not rebuild a script.

what's inside

Services and prices

A current catalog the AI quotes conditions from: no stale data.

Rules and limits

What may be promised a client and what may not: the AI stays inside the frame.

Documents as a source

Manuals, policies and descriptions plug into the base, and the AI answers from them.

Stop-topics

Questions the AI does not answer itself, but calls a human for at once.

Answers from facts

The AI leans on the base and the client card, not on a network guesses.

Handoff to a human

The operator gets the dialogue with the handoff reason, a summary and the client history.

Updates without a developer

Conditions changed: you edit a record in your own words. This is knowledge base software you keep current yourself.

Answer debugging

An inaccuracy is fixed in one step and checked on a test dialogue.

One base, every channel

WhatsApp, Telegram, the site and voice all answer from the same facts.

Tone of voice

One brand style, from strict to friendly.

what changes

What changes when the AI answers from your facts

Repeat questions stop reaching people: the schedule, terms, availability and booking steps the bot closes itself, day and night. Admins handle the hard cases instead of dictating the same thing in circles. And the answers are identical everywhere the business appears: the site, WhatsApp and Telegram speak with one set of facts.

Changes in the business stop breaking support. A service appeared or the schedule shifted: edit one record, and the fresh terms reach every conversation without waiting or retraining. No need to gather the admins for a briefing and hunt down whoever still answers the old way. And sensitive topics go to a live person along with the question history, so the client is never left alone with a machine at the important moment.

how much it costs

$59 /mo

included in plans from Start

The knowledge base and first-line support are included from the Start plan: extended bases, complex handoff rules and helpdesk integrations move up into Business and above.

frequently asked questions

Will the AI make up answers?

No, the script is built around the knowledge base, the rules and handoff to a human. If a question is outside the base, the AI hands the dialogue to an operator instead of inventing.

Can I use it in medicine or education?

Yes, but with careful setup: factual limits, a ban on dangerous advice, handoff to a specialist and separate safety rules.

How do I update the knowledge?

Through the settings and the debugger: you add a rule or correct an answer in your own words, then check the script on a test dialogue. No developer needed.

What if the answer is not in the base?

The AI will not guess: it honestly says it will check, and hands the question to a human with a summary and the dialogue history.

Can I link it to our CRM or helpdesk?

Yes, integration with a CRM and a helpdesk attaches as an extension. Non-standard integrations we work through at launch.